Commit 31775c5a authored by Matthew Dempsky's avatar Matthew Dempsky Committed by Ian Lance Taylor

cmd/cgo: update code and docs to reflect post-6c world

The gc toolchain no longer includes a C compiler, so mentions of "6c"
can be removed or replaced by 6g as appropriate.  Similarly, some cgo
functions that previously generated C source output no longer need to.

Change-Id: I1ae6b02630cff9eaadeae6f3176c0c7824e8fbe5
Reviewed-on: https://go-review.googlesource.com/2391Reviewed-by: 's avatarIan Lance Taylor <iant@golang.org>
parent 3ef39472
......@@ -213,10 +213,10 @@ placed in preambles in other files, or in C source files.
Using cgo directly
Usage:
go tool cgo [cgo options] [-- compiler options] file.go
go tool cgo [cgo options] [-- compiler options] gofiles...
Cgo transforms the input file.go into four output files: two Go source
files, a C file for 6c (or 8c or 5c), and a C file for gcc.
Cgo transforms the specified input Go source files into several output
Go and C source files.
The compiler options are passed through uninterpreted when
invoking the C compiler to compile the C parts of the package.
......@@ -229,6 +229,8 @@ The following options are available when running cgo directly:
build when building a cgo package.
-dynout file
Write -dynimport output to file.
-dynpackage package
Set Go package for -dynimport output.
-dynlinker
Write dynamic linker as part of -dynimport output.
-godefs
......@@ -375,9 +377,9 @@ the translation process.
Translating Go
[The rest of this comment refers to 6g and 6c, the Go and C compilers
that are part of the amd64 port of the gc Go toolchain. Everything here
applies to another architecture's compilers as well.]
[The rest of this comment refers to 6g, the Go compiler that is part
of the amd64 port of the gc Go toolchain. Everything here applies to
another architecture's compilers as well.]
Given the input Go files x.go and y.go, cgo generates these source
files:
......@@ -385,44 +387,41 @@ files:
x.cgo1.go # for 6g
y.cgo1.go # for 6g
_cgo_gotypes.go # for 6g
_cgo_defun.c # for 6c
_cgo_import.go # for 6g (if -dynout _cgo_import.go)
x.cgo2.c # for gcc
y.cgo2.c # for gcc
_cgo_defun.c # for gcc (if -gccgo)
_cgo_export.c # for gcc
_cgo_export.h # for gcc
_cgo_main.c # for gcc
_cgo_flags # for alternative build tools
The file x.cgo1.go is a copy of x.go with the import "C" removed and
references to C.xxx replaced with names like _Cfunc_xxx or _Ctype_xxx.
The definitions of those identifiers, written as Go functions, types,
or variables, are provided in _cgo_gotypes.go.
Here is a _cgo_gotypes.go containing definitions for C.flush (provided
in the preamble) and C.puts (from stdio):
Here is a _cgo_gotypes.go containing definitions for needed C types:
type _Ctype_char int8
type _Ctype_int int32
type _Ctype_void [0]byte
func _Cfunc_CString(string) *_Ctype_char
func _Cfunc_flush() _Ctype_void
func _Cfunc_puts(*_Ctype_char) _Ctype_int
For functions, cgo only writes an external declaration in the Go
output. The implementation is in a combination of C for 6c (meaning
any gc-toolchain compiler) and C for gcc.
The 6c file contains the definitions of the functions. They all have
similar bodies that invoke runtime·cgocall to make a switch from the
Go runtime world to the system C (GCC-based) world.
The _cgo_gotypes.go file also contains the definitions of the
functions. They all have similar bodies that invoke runtime·cgocall
to make a switch from the Go runtime world to the system C (GCC-based)
world.
For example, here is the definition of _Cfunc_puts:
void _cgo_be59f0f25121_Cfunc_puts(void*);
//go:cgo_import_static _cgo_be59f0f25121_Cfunc_puts
//go:linkname __cgofn__cgo_be59f0f25121_Cfunc_puts _cgo_be59f0f25121_Cfunc_puts
var __cgofn__cgo_be59f0f25121_Cfunc_puts byte
var _cgo_be59f0f25121_Cfunc_puts = unsafe.Pointer(&__cgofn__cgo_be59f0f25121_Cfunc_puts)
void
·_Cfunc_puts(struct{uint8 x[1];}p)
{
runtime·cgocall(_cgo_be59f0f25121_Cfunc_puts, &p);
func _Cfunc_puts(p0 *_Ctype_char) (r1 _Ctype_int) {
_cgo_runtime_cgocall_errno(_cgo_be59f0f25121_Cfunc_puts, uintptr(unsafe.Pointer(&p0)))
return
}
The hexadecimal number is a hash of cgo's input, chosen to be
......@@ -468,23 +467,21 @@ code generated for gcc. The build process links this stub, along with
_cgo_export.c and *.cgo2.c, into a dynamic executable and then lets
cgo examine the executable. Cgo records the list of shared library
references and resolved names and writes them into a new file
_cgo_import.c, which looks like:
_cgo_import.go, which looks like:
#pragma cgo_dynamic_linker "/lib64/ld-linux-x86-64.so.2"
#pragma cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
#pragma cgo_import_dynamic __libc_start_main __libc_start_main#GLIBC_2.2.5 "libc.so.6"
#pragma cgo_import_dynamic stdout stdout#GLIBC_2.2.5 "libc.so.6"
#pragma cgo_import_dynamic fflush fflush#GLIBC_2.2.5 "libc.so.6"
#pragma cgo_import_dynamic _ _ "libpthread.so.0"
#pragma cgo_import_dynamic _ _ "libc.so.6"
//go:cgo_dynamic_linker "/lib64/ld-linux-x86-64.so.2"
//go:cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
//go:cgo_import_dynamic __libc_start_main __libc_start_main#GLIBC_2.2.5 "libc.so.6"
//go:cgo_import_dynamic stdout stdout#GLIBC_2.2.5 "libc.so.6"
//go:cgo_import_dynamic fflush fflush#GLIBC_2.2.5 "libc.so.6"
//go:cgo_import_dynamic _ _ "libpthread.so.0"
//go:cgo_import_dynamic _ _ "libc.so.6"
In the end, the compiled Go package, which will eventually be
presented to 6l as part of a larger program, contains:
_go_.6 # 6g-compiled object for _cgo_gotypes.go *.cgo1.go
_cgo_defun.6 # 6c-compiled object for _cgo_defun.c
_go_.6 # 6g-compiled object for _cgo_gotypes.go, _cgo_import.go, *.cgo1.go
_all.o # gcc-compiled object for _cgo_export.c, *.cgo2.c
_cgo_import.6 # 6c-compiled object for _cgo_import.c
The final program will be a dynamic executable, so that 6l can avoid
needing to process arbitrary .o files. It only needs to process the .o
......@@ -508,20 +505,21 @@ Runtime
When using cgo, Go must not assume that it owns all details of the
process. In particular it needs to coordinate with C in the use of
threads and thread-local storage. The runtime package, in its own
(6c-compiled) C code, declares a few uninitialized (default bss)
threads and thread-local storage. The runtime package declares a few
variables:
bool runtime·iscgo;
void (*libcgo_thread_start)(void*);
void (*initcgo)(G*);
var (
iscgo bool
_cgo_init unsafe.Pointer
_cgo_thread_start unsafe.Pointer
)
Any package using cgo imports "runtime/cgo", which provides
initializations for these variables. It sets iscgo to 1, initcgo to a
gcc-compiled function that can be called early during program startup,
and libcgo_thread_start to a gcc-compiled function that can be used to
create a new thread, in place of the runtime's usual direct system
calls.
initializations for these variables. It sets iscgo to true, _cgo_init
to a gcc-compiled function that can be called early during program
startup, and _cgo_thread_start to a gcc-compiled function that can be
used to create a new thread, in place of the runtime's usual direct
system calls.
Internal and External Linking
......@@ -534,12 +532,12 @@ code can only be used as a dynamic library). On the other hand, when
using internal linking, 6l can generate Go binaries by itself.
In order to allow linking arbitrary object files without requiring
dynamic libraries, cgo will soon support an "external" linking mode
too. In external linking mode, 6l does not process any host object
files. Instead, it collects all the Go code and writes a single go.o
object file containing it. Then it invokes the host linker (usually
gcc) to combine the go.o object file and any supporting non-Go code
into a final executable. External linking avoids the dynamic library
dynamic libraries, cgo supports an "external" linking mode too. In
external linking mode, 6l does not process any host object files.
Instead, it collects all the Go code and writes a single go.o object
file containing it. Then it invokes the host linker (usually gcc) to
combine the go.o object file and any supporting non-Go code into a
final executable. External linking avoids the dynamic library
requirement but introduces a requirement that the host linker be
present to create such a binary.
......@@ -567,13 +565,13 @@ to be made when linking the final binary.
Linking Directives
In either linking mode, package-specific directives must be passed
through to 6l. These are communicated by writing #pragma directives
in a C source file compiled by 6c. The directives are copied into the .6 object file
and then processed by the linker.
through to 6l. These are communicated by writing //go: directives in a
Go source file compiled by 6g. The directives are copied into the .6
object file and then processed by the linker.
The directives are:
#pragma cgo_import_dynamic <local> [<remote> ["<library>"]]
//go:cgo_import_dynamic <local> [<remote> ["<library>"]]
In internal linking mode, allow an unresolved reference to
<local>, assuming it will be resolved by a dynamic library
......@@ -584,9 +582,9 @@ The directives are:
In the <remote>, # or @ can be used to introduce a symbol version.
Examples:
#pragma cgo_import_dynamic puts
#pragma cgo_import_dynamic puts puts#GLIBC_2.2.5
#pragma cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
//go:cgo_import_dynamic puts
//go:cgo_import_dynamic puts puts#GLIBC_2.2.5
//go:cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
A side effect of the cgo_import_dynamic directive with a
library is to make the final binary depend on that dynamic
......@@ -594,12 +592,12 @@ The directives are:
symbols, use _ for local and remote.
Example:
#pragma cgo_import_dynamic _ _ "libc.so.6"
//go:cgo_import_dynamic _ _ "libc.so.6"
For compatibility with current versions of SWIG,
#pragma dynimport is an alias for #pragma cgo_import_dynamic.
#pragma dynimport is an alias for //go:cgo_import_dynamic.
#pragma cgo_dynamic_linker "<path>"
//go:cgo_dynamic_linker "<path>"
In internal linking mode, use "<path>" as the dynamic linker
in the final binary. This directive is only needed from one
......@@ -607,9 +605,9 @@ The directives are:
supplied by runtime/cgo.
Example:
#pragma cgo_dynamic_linker "/lib/ld-linux.so.2"
//go:cgo_dynamic_linker "/lib/ld-linux.so.2"
#pragma cgo_export_dynamic <local> <remote>
//go:cgo_export_dynamic <local> <remote>
In internal linking mode, put the Go symbol
named <local> into the program's exported symbol table as
......@@ -618,9 +616,9 @@ The directives are:
to share Go's data.
For compatibility with current versions of SWIG,
#pragma dynexport is an alias for #pragma cgo_export_dynamic.
#pragma dynexport is an alias for //go:cgo_export_dynamic.
#pragma cgo_import_static <local>
//go:cgo_import_static <local>
In external linking mode, allow unresolved references to
<local> in the go.o object file prepared for the host linker,
......@@ -628,9 +626,9 @@ The directives are:
other object files that will be linked with go.o.
Example:
#pragma cgo_import_static puts_wrapper
//go:cgo_import_static puts_wrapper
#pragma cgo_export_static <local> <remote>
//go:cgo_export_static <local> <remote>
In external linking mode, put the Go symbol
named <local> into the program's exported symbol table as
......@@ -638,15 +636,15 @@ The directives are:
mechanism makes it possible for C code to call back into Go or
to share Go's data.
#pragma cgo_ldflag "<arg>"
//go:cgo_ldflag "<arg>"
In external linking mode, invoke the host linker (usually gcc)
with "<arg>" as a command-line argument following the .o files.
Note that the arguments are for "gcc", not "ld".
Example:
#pragma cgo_ldflag "-lpthread"
#pragma cgo_ldflag "-L/usr/local/sqlite3/lib"
//go:cgo_ldflag "-lpthread"
//go:cgo_ldflag "-L/usr/local/sqlite3/lib"
A package compiled with cgo will include directives for both
internal and external linking; the linker will select the appropriate
......@@ -659,22 +657,18 @@ The following code will be generated by cgo:
// compiled by 6g
type _Ctype_double float64
func _Cfunc_sin(_Ctype_double) _Ctype_double
// compiled by 6c
#pragma cgo_import_dynamic sin sin#GLIBC_2.2.5 "libm.so.6"
//go:cgo_ldflag "-lm"
#pragma cgo_import_static _cgo_gcc_Cfunc_sin
#pragma cgo_ldflag "-lm"
type _Ctype_double float64
void _cgo_gcc_Cfunc_sin(void*);
//go:cgo_import_static _cgo_gcc_Cfunc_sin
//go:linkname __cgo_gcc_Cfunc_sin _cgo_gcc_Cfunc_sin
var __cgo_gcc_Cfunc_sin byte
var _cgo_gcc_Cfunc_sin = unsafe.Pointer(&__cgo_gcc_Cfunc_sin)
void
·_Cfunc_sin(struct{uint8 x[16];}p)
{
runtime·cgocall(_cgo_gcc_Cfunc_sin, &p);
func _Cfunc_sin(p0 _Ctype_double) (r1 _Ctype_double) {
_cgo_runtime_cgocall_errno(_cgo_gcc_Cfunc_sin, uintptr(unsafe.Pointer(&p0)))
return
}
// compiled by gcc, into foo.cgo2.o
......@@ -694,8 +688,8 @@ using the internal or external mode. If other packages are compiled in
"external only" mode, then the final link will be an external one.
Otherwise the link will be an internal one.
The directives in the 6c-compiled file are used according to the kind
of final link used.
The linking directives are used according to the kind of final link
used.
In internal mode, 6l itself processes all the host object files, in
particular foo.cgo2.o. To do so, it uses the cgo_import_dynamic and
......@@ -706,10 +700,10 @@ symbol sin with version GLIBC_2.2.5 from the dynamic library
runtime dynamic linker.
In external mode, 6l does not process any host object files, in
particular foo.cgo2.o. It links together the 6g- and 6c-generated
object files, along with any other Go code, into a go.o file. While
doing that, 6l will discover that there is no definition for
_cgo_gcc_Cfunc_sin, referred to by the 6c-compiled source file. This
particular foo.cgo2.o. It links together the 6g-generated object
files, along with any other Go code, into a go.o file. While doing
that, 6l will discover that there is no definition for
_cgo_gcc_Cfunc_sin, referred to by the 6g-compiled source file. This
is okay, because 6l also processes the cgo_import_static directive and
knows that _cgo_gcc_Cfunc_sin is expected to be supplied by a host
object file, so 6l does not treat the missing symbol as an error when
......
......@@ -150,9 +150,9 @@ var cPrefix string
var fset = token.NewFileSet()
var dynobj = flag.String("dynimport", "", "if non-empty, print dynamic import data for that file")
var dynout = flag.String("dynout", "", "write -dynobj output to this file")
var dynpackage = flag.String("dynpackage", "main", "set Go package for dynobj output")
var dynlinker = flag.Bool("dynlinker", false, "record dynamic linker information in dynimport mode")
var dynout = flag.String("dynout", "", "write -dynimport output to this file")
var dynpackage = flag.String("dynpackage", "main", "set Go package for -dynimport output")
var dynlinker = flag.Bool("dynlinker", false, "record dynamic linker information in -dynimport mode")
// These flags are for bootstrapping a new Go implementation,
// to generate Go and C headers that match the data layout and
......
......@@ -21,8 +21,8 @@ import (
var conf = printer.Config{Mode: printer.SourcePos, Tabwidth: 8}
// writeDefs creates output files to be compiled by 6g, 6c, and gcc.
// (The comments here say 6g and 6c but the code applies to the 8 and 5 tools too.)
// writeDefs creates output files to be compiled by 6g and gcc.
// (The comments here say 6g but the code applies to the 8 and 5 tools too.)
func (p *Package) writeDefs() {
var fgo2, fc io.Writer
f := creat(*objDir + "_cgo_gotypes.go")
......@@ -159,14 +159,14 @@ func (p *Package) writeDefs() {
for _, key := range nameKeys(p.Name) {
n := p.Name[key]
if n.FuncType != nil {
p.writeDefsFunc(fc, fgo2, n)
p.writeDefsFunc(fgo2, n)
}
}
if *gccgo {
p.writeGccgoExports(fgo2, fc, fm)
p.writeGccgoExports(fgo2, fm)
} else {
p.writeExports(fgo2, fc, fm)
p.writeExports(fgo2, fm)
}
init := gccgoInit.String()
......@@ -258,10 +258,10 @@ func dynimport(obj string) {
fatalf("cannot parse %s as ELF, Mach-O or PE", obj)
}
// Construct a gcc struct matching the 6c argument frame.
// Construct a gcc struct matching the 6g argument frame.
// Assumes that in gcc, char is 1 byte, short 2 bytes, int 4 bytes, long long 8 bytes.
// These assumptions are checked by the gccProlog.
// Also assumes that 6c convention is to word-align the
// Also assumes that 6g convention is to word-align the
// input and output parameters.
func (p *Package) structType(n *Name) (string, int64) {
var buf bytes.Buffer
......@@ -310,7 +310,7 @@ func (p *Package) structType(n *Name) (string, int64) {
return buf.String(), off
}
func (p *Package) writeDefsFunc(fc, fgo2 io.Writer, n *Name) {
func (p *Package) writeDefsFunc(fgo2 io.Writer, n *Name) {
name := n.Go
gtype := n.FuncType.Go
void := gtype.Results == nil || len(gtype.Results.List) == 0
......@@ -442,7 +442,7 @@ func (p *Package) writeDefsFunc(fc, fgo2 io.Writer, n *Name) {
}
// writeOutput creates stubs for a specific source file to be compiled by 6g
// (The comments here say 6g and 6c but the code applies to the 8 and 5 tools too.)
// (The comments here say 6g but the code applies to the 8 and 5 tools too.)
func (p *Package) writeOutput(f *File, srcfile string) {
base := srcfile
if strings.HasSuffix(base, ".go") {
......@@ -459,7 +459,7 @@ func (p *Package) writeOutput(f *File, srcfile string) {
fmt.Fprintf(fgo1, "// Created by cgo - DO NOT EDIT\n\n")
conf.Fprint(fgo1, fset, f.AST)
// While we process the vars and funcs, also write 6c and gcc output.
// While we process the vars and funcs, also write gcc output.
// Gcc output starts with the preamble.
fmt.Fprintf(fgcc, "%s\n", f.Preamble)
fmt.Fprintf(fgcc, "%s\n", gccProlog)
......@@ -521,7 +521,7 @@ func (p *Package) writeOutputFunc(fgcc *os.File, n *Name) {
if n.AddError {
fmt.Fprintf(fgcc, "\terrno = 0;\n")
}
// We're trying to write a gcc struct that matches 6c/8c/5c's layout.
// We're trying to write a gcc struct that matches 6g's layout.
// Use packed attribute to force no padding in this struct in case
// gcc has different packing requirements.
fmt.Fprintf(fgcc, "\t%s %v *a = v;\n", ctype, p.packedAttribute())
......@@ -617,8 +617,8 @@ func (p *Package) writeGccgoOutputFunc(fgcc *os.File, n *Name) {
}
// packedAttribute returns host compiler struct attribute that will be
// used to match 6c/8c/5c's struct layout. For example, on 386 Windows,
// gcc wants to 8-align int64s, but 8c does not.
// used to match 6g's struct layout. For example, on 386 Windows,
// gcc wants to 8-align int64s, but 8g does not.
// Use __gcc_struct__ to work around http://gcc.gnu.org/PR52991 on x86,
// and http://golang.org/issue/5603.
func (p *Package) packedAttribute() string {
......@@ -631,7 +631,7 @@ func (p *Package) packedAttribute() string {
// Write out the various stubs we need to support functions exported
// from Go so that they are callable from C.
func (p *Package) writeExports(fgo2, fc, fm io.Writer) {
func (p *Package) writeExports(fgo2, fm io.Writer) {
fgcc := creat(*objDir + "_cgo_export.c")
fgcch := creat(*objDir + "_cgo_export.h")
......@@ -647,7 +647,7 @@ func (p *Package) writeExports(fgo2, fc, fm io.Writer) {
for _, exp := range p.ExpFunc {
fn := exp.Func
// Construct a gcc struct matching the 6c argument and
// Construct a gcc struct matching the 6g argument and
// result frame. The gcc struct will be compiled with
// __attribute__((packed)) so all padding must be accounted
// for explicitly.
......@@ -763,7 +763,7 @@ func (p *Package) writeExports(fgo2, fc, fm io.Writer) {
}
fmt.Fprintf(fgcc, "}\n")
// Build the wrapper function compiled by 6c/8c
// Build the wrapper function compiled by 6g.
goname := exp.Func.Name.Name
if fn.Recv != nil {
goname = "_cgoexpwrap" + cPrefix + "_" + fn.Recv.List[0].Names[0].Name + "_" + goname
......@@ -822,7 +822,7 @@ func (p *Package) writeExports(fgo2, fc, fm io.Writer) {
}
// Write out the C header allowing C code to call exported gccgo functions.
func (p *Package) writeGccgoExports(fgo2, fc, fm io.Writer) {
func (p *Package) writeGccgoExports(fgo2, fm io.Writer) {
fgcc := creat(*objDir + "_cgo_export.c")
fgcch := creat(*objDir + "_cgo_export.h")
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment